detecto

Why we included this project

Detecto is a thin Python wrapper around PyTorch's Faster R-CNN object detection models, and the whole point is that you get a working detector in a few lines of code. Instead of hand-writing training loops, data loaders, and inference plumbing, you load a labeled dataset (XML or CSV), fine-tune a pretrained model on your own classes, and run predictions on still images, video files, or a live webcam feed. Since it sits directly on torchvision, you can swap in lighter backbones like MobileNet for constrained deployments and still drop down to the underlying PyTorch model when you need finer control. For developers who want a quick, readable starting point for custom object detection, or a teaching-friendly layer over PyTorch, it's a low-ceremony option with clear documentation.